[llvm] [AMDGPU] Add AMDPAL support for llvm.debugtrap (PR #219179)

[Matt Arsenault via llvm-commits]

Alexander =?utf-8?q?Hück?= <alexander.huck at amd.com>,
Alexander =?utf-8?q?Hück?= <alexander.huck at amd.com>,
Alexander =?utf-8?q?Hück?= <alexander.huck at amd.com>,
Alexander =?utf-8?q?Hück?= <alexander.huck at amd.com>,
Alexander =?utf-8?q?Hück?= <alexander.huck at amd.com>
Message-ID:
In-Reply-To: <llvm.org/llvm/llvm-project/pull/219179 at github.com>


================
@@ -63,13 +64,28 @@ template <class T> static void forEachCall(Function &Intrin, T Callback) {
 
 } // anonymous namespace
 
+bool AMDGPULowerIntrinsicsImpl::visitIsDebuggingEnabled(IntrinsicInst &I) {
+  const GCNSubtarget &ST = TM.getSubtarget<GCNSubtarget>(*I.getFunction());
+  if (ST.hasCDBGSysOrUserBranch())
+    return false;
+
+  I.replaceAllUsesWith(ConstantInt::getFalse(I.getContext()));
+  I.eraseFromParent();
+  return true;
----------------
arsenm wrote:

This doesn't need IR pass handling (if anything, based on the hacking up the legalizer, it would make more sense to handle the legal case here)

https://github.com/llvm/llvm-project/pull/219179